Old people daily behavior pattern analysis and risk early warning method based on edge computing

Abstract

This invention relates to the field of smart elderly care technology, and discloses a method for analyzing and warning of daily behavior patterns of the elderly based on edge computing. The method includes: continuously collecting multi-source behavioral data of the elderly at edge computing nodes deployed within their living spaces; identifying behavioral trajectory patterns, constructing an activity space map of the elderly using these patterns, and analyzing the activity patterns based on the activity space map; analyzing the elderly's device usage characteristics, constructing habitual behavior chains based on these characteristics, and analyzing their daily routine characteristics based on these habitual behavior chains; and analyzing the deviation of the elderly's daily behavior using these activity and routine characteristics, and constructing a behavioral risk warning report based on the deviation. This invention can achieve comprehensive perception of the daily behavior of the elderly and improve the accuracy of behavioral warnings.

Description

Technical Field

[0001] This invention relates to a method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing, belonging to the field of smart elderly care technology. Background Technology

[0002] With the accelerating aging of the global population, improving the quality of life for the elderly and addressing the social pressures of elder care has become a pressing issue. Continuous monitoring and intelligent analysis of the daily behavioral data of the elderly can promptly identify abnormal changes in behavioral patterns, thereby providing more precise and timely safety and security services. This is also of significant value in improving the quality of elder care services and reducing the incidence of accidents.

[0003] Currently, the analysis and risk warning of daily behavior in the elderly largely rely on centralized cloud processing or single sensor data collection. For example, indoor positioning sensors collect the elderly's activity trajectories and upload them to the cloud for analysis, or device interaction data is used to identify simple behavioral anomalies. These methods can initially identify some behavioral deviations in the elderly, such as an abnormally reduced activity range or the absence of fixed behaviors. However, in traditional solutions, cloud processing requires the remote transmission of large amounts of data, which can easily lead to delayed warnings due to network latency. Furthermore, the narrow dimensions of single data collection make it difficult to comprehensively reconstruct behavioral scenarios, ultimately resulting in insufficient accuracy in analyzing elderly behavior patterns and inadequate timeliness of risk warnings. Summary of the Invention

[0004] This invention provides a method for analyzing and warning of daily behavior patterns of the elderly based on edge computing. Its main purpose is to achieve comprehensive perception of the daily behavior of the elderly and improve the accuracy of behavior warning.

[0005] To achieve the above objectives, the present invention provides a method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing, comprising:

[0006] At edge computing nodes deployed in the living spaces of the elderly, multi-source behavioral data of the elderly is continuously collected. The multi-source behavioral data includes spatial positioning signals and device interaction signals. Using the spatial positioning signal, the behavioral trajectory pattern of the elderly is identified. The behavioral trajectory pattern includes dynamic movement trajectory and static hotspot. Using the behavioral trajectory pattern, an activity space map of the elderly is constructed. Based on the activity space map, the activity pattern characteristics of the elderly are analyzed. Based on the device interaction signals, analyze the device usage characteristics of the elderly, construct the habitual behavior chain of the elderly based on the device usage characteristics, and analyze the daily routine characteristics of the elderly based on the habitual behavior chain. By utilizing the activity pattern characteristics and the daily routine characteristics, the behavioral deviation degree of the elderly is analyzed, and a behavioral risk warning report of the elderly is constructed based on the behavioral deviation degree.

[0007] Optionally, the behavioral deviation of the elderly's daily behavior can be analyzed using the activity pattern characteristics and the daily routine characteristics, including: Using the aforementioned activity patterns, the intensity of abnormal fluctuations in the daily behavior of the elderly over time was analyzed. The intensity of the fluctuation anomaly is used as the first deviation factor of the elderly person's behavior; Using the aforementioned characteristics of daily routines, the degree of disruption in the periodicity and stability of the daily routines of the elderly was analyzed, resulting in a second deviation factor; Using the first deviation factor and the second deviation factor, calculate the comprehensive behavioral deviation coefficient of the elderly person's daily behavior; Based on the comprehensive behavioral deviation coefficient, the behavioral deviation degree of the elderly person's daily behavior is determined.

[0008] Optionally, using the first deviation factor and the second deviation factor, a comprehensive behavioral deviation coefficient for the daily behavior of the elderly is calculated, including: The first deviation factor and the second deviation factor are nonlinearly corrected to obtain the first linear factor and the second linear factor. The first linear factor and the second linear factor are adjusted for differences to obtain the first enhancement factor and the second enhancement factor. Based on the first enhancement factor and the second enhancement factor, the comprehensive behavioral deviation coefficient of the elderly person's daily behavior is calculated.

[0009] Optionally, using the spatial positioning signal to identify the behavioral trajectory patterns of the elderly includes: Trajectory point clustering analysis is performed on the spatial positioning signal to identify the static hotspots and dwell time of the elderly using the decomposition results of the trajectory point clustering analysis; Identify the spatial positioning sequence of the spatial positioning signal, and use the spatial positioning sequence to construct the dynamic movement trajectory of the elderly person; Based on the static hotspots, the duration of stay, and the dynamic movement trajectory, the behavioral trajectory patterns of the elderly are identified.

[0010] Optionally, based on the activity space map, the activity pattern characteristics of the elderly are analyzed, including: Query the historical activity data of the elderly person to construct a baseline activity map of the elderly person using the historical activity data; Extract spatial distribution indicators of the same dimension from the activity space map and the benchmark activity map to obtain a comparison indicator set; Based on the comparison index set, the activity space map is annotated with abnormal regions to obtain annotated activity regions; The activity patterns of the elderly can be identified by using the marked activity areas.

[0011] Optionally, based on the habitual behavior chain, the daily routine characteristics of the elderly are analyzed, including: The habitual behavior chain is filtered for key time periods to obtain the core behavior time period group; Based on the core behavior time period groups, identify the time distribution pattern of the habitual behavior chain; By utilizing the aforementioned time distribution patterns, key behavioral periods of the elderly can be identified, thus revealing critical daily routines. By utilizing the aforementioned key sleep periods, the sleep patterns of the elderly can be identified.

[0012] Optionally, based on the device interaction signals, the device usage characteristics of the elderly person are analyzed, including: Using the device interaction signals, a device usage table for the elderly is constructed, which includes device usage events and usage timestamps; Based on the device usage table, the frequency of use and average usage time of each device used by the elderly were analyzed; Based on the frequency of use and the average duration of use, the device usage characteristics of the elderly are identified.

[0013] Optionally, using the behavioral trajectory pattern, an activity space map of the elderly person is constructed, including: Identify high-frequency static hotspots and mainstream dynamic movement paths in the behavioral trajectory patterns; Using the aforementioned mainstream dynamic movement paths, a spatial activity network model for the elderly is constructed. The high-frequency static hotspots are spatiotemporally superimposed with the spatial activity network model to obtain an activity space map.

[0014] Optionally, based on the device usage characteristics, a habitual behavior chain of the elderly person is constructed, including: The device usage events in the device usage characteristics are sorted by timestamps to obtain a device usage event sequence; The device usage event sequence is subjected to association rule mining to identify device usage events that are frequently and continuously triggered in time sequence, thereby obtaining frequent sequence patterns; Based on the frequent sequence patterns, the habitual behavior chain of the elderly is constructed.

[0015] Optionally, based on the behavioral deviation, a behavioral risk warning report for the elderly is constructed, including: Based on the behavioral deviation, a behavioral risk level for the elderly is generated; Based on the behavioral risk level, construct a behavioral risk early warning report for the elderly.

[0016] To address the aforementioned problems, this invention also provides an edge computing-based system for analyzing and warning of daily behavior patterns in the elderly, the system comprising: The behavior signal acquisition module is used to continuously collect multi-source behavior data of the elderly at the edge computing node deployed in the living space of the elderly. The multi-source behavior data includes spatial positioning signals and device interaction signals. The activity pattern analysis module is used to identify the behavioral trajectory patterns of the elderly using the spatial positioning signal. The behavioral trajectory patterns include dynamic movement trajectories and static hotspots. The module uses the behavioral trajectory patterns to construct the activity space map of the elderly and analyzes the activity pattern characteristics of the elderly based on the activity space map. The daily routine analysis module is used to analyze the device usage characteristics of the elderly based on the device interaction signals, construct the habitual behavior chain of the elderly based on the device usage characteristics, and analyze the daily routine characteristics of the elderly based on the habitual behavior chain. The abnormality warning module is used to analyze the behavioral deviation of the elderly's daily behavior by utilizing the activity pattern characteristics and the daily routine characteristics, and to construct a behavioral risk warning report for the elderly based on the behavioral deviation.

[0017] Compared to the problems described in the background technology, this invention first continuously collects multi-source behavioral data by deploying edge computing nodes in the living spaces of the elderly. Spatial positioning signals capture the physical movement trajectories of the elderly, while device interaction signals reflect their interactions with everyday devices. The combination of these two comprehensively covers the spatial activities and tool usage scenarios in the lives of the elderly, providing complete data support for accurate identification of behavioral patterns. Next, this invention identifies behavioral trajectory patterns including dynamic movement trajectories and static dwelling hotspots to accurately reconstruct the specific activity paths and high-frequency dwelling areas of the elderly in their living spaces. Based on this, an activity space map is constructed, transforming abstract trajectory data into a visualized, structured spatial model that intuitively presents the usage of each area. The invention analyzes the frequency, duration, and regional relationships of elderly individuals' use of various devices. It quantifies their usage frequency, duration, and time preferences by examining device usage characteristics, clearly revealing their tool-dependent habits. Based on these characteristics, a habitual behavior chain is constructed, logically linking isolated device usage events over time, providing a comprehensive reference framework for abnormal deviation analysis. Finally, the invention analyzes behavioral deviation by integrating activity and sleep patterns, and constructs a behavioral risk warning report to visually present the risk of abnormal behavior in the elderly. For example, if an elderly person's wake-up time deviates from the norm by 2 hours and the medication reminder device is not used beyond the acceptable range, it provides caregivers with a clear focus and timely understanding of potential risks. Therefore, this invention enables comprehensive perception of the daily behavior of the elderly and improves the accuracy of behavioral warnings. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing, according to an embodiment of the present invention. Figure 2 This is a schematic diagram of a module for implementing edge computing-based analysis of daily behavior patterns and risk warning of the elderly, provided in an embodiment of the present invention.

[0019] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0021] This application provides a method for analyzing and warning of daily behavior patterns and risks of the elderly based on edge computing. The executing entity of this method includes, but is not limited to, at least one electronic device that can be configured to execute the method provided in this application, such as a server or a terminal. In other words, the edge computing-based system for analyzing and warning of daily behavior patterns and risks of the elderly can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0022] Reference Figure 1 The diagram shown is a flowchart illustrating a method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing, according to an embodiment of the present invention. In this embodiment, the method includes: S1. At the edge computing node deployed in the living space of the elderly, multi-source behavioral data of the elderly are continuously collected. The multi-source behavioral data includes spatial positioning signals and device interaction signals.

[0023] This invention, through edge computing nodes deployed in the living spaces of the elderly, continuously collects multi-source behavioral data of the elderly. In the elderly's living scenarios, the edge computing nodes can obtain real-time multi-source raw data of spatial positioning and device interaction in real time, providing a real and continuous foundation for subsequent behavioral pattern analysis. For example, the edge nodes at home can continuously collect the elderly's residence signals in the living room and the interaction records of using smart speakers.

[0024] The multi-source behavioral data includes spatial positioning signals and device interaction signals. The spatial positioning signals refer to continuous spatiotemporal data reflecting the real-time location, movement speed, movement direction, and area of ​​residence of the elderly in their living space. The device interaction signals refer to data reflecting changes in the status of the elderly using specific devices or performing specific daily activities.

[0025] In detail, edge computing nodes and supporting sensing devices can be deployed in key living areas for the elderly, such as bedrooms, living rooms, small parks, and senior activity centers, to capture multi-source data such as spatial positioning and device interaction in real time and continuously upload and store them.

[0026] S2. Using the spatial positioning signal, identify the behavioral trajectory pattern of the elderly, which includes dynamic movement trajectory and static hotspot. Using the behavioral trajectory pattern, construct the activity space map of the elderly, and analyze the activity pattern characteristics of the elderly based on the activity space map.

[0027] This invention utilizes the spatial positioning signal to identify the behavioral trajectory patterns of the elderly, transforming the original spatial positioning signal into interpretable behavioral trajectory information. This accurately presents the elderly's movement path and high-frequency residence areas in their living space, such as identifying the elderly's morning residence hotspots in the kitchen and their dynamic movement trajectory between the living room and the balcony.

[0028] As an embodiment of the present invention, the spatial positioning signal is used to identify the behavioral trajectory patterns of the elderly, including: Trajectory point clustering analysis is performed on the spatial positioning signal to identify the static hotspots and dwell time of the elderly using the decomposition results of the trajectory point clustering analysis; Identify the spatial positioning sequence of the spatial positioning signal, and use the spatial positioning sequence to construct the dynamic movement trajectory of the elderly person; Based on the static hotspots, the duration of stay, and the dynamic movement trajectory, the behavioral trajectory patterns of the elderly are identified.

[0029] In detail, a density-based clustering algorithm can be used to automatically group densely distributed points in spatial positioning signals into multiple clusters, each cluster corresponding to a static hotspot. Then, based on the timestamp sequence of points within each cluster, the time difference between the first and last points is calculated to determine the specific dwell time of the elderly person at each hotspot. Adjacent positioning points are connected according to time sequence to generate a coherent movement trajectory. Simultaneously, the ratio of displacement to time interval between adjacent points is extracted from the trajectory sequence to calculate the movement speed, and the rate of change of direction angle is analyzed to capture subtle features such as acceleration and turning during movement, thus obtaining a dynamic movement trajectory. Finally, based on the identified static hotspots, dwell times, and dynamic movement trajectories, a spatiotemporal correlation model is used to integrate the hotspots and trajectories, analyzing the movement frequency between hotspots and the repetition patterns of the trajectories. For example, by combining the distribution of dwell time and the similarity of movement paths, periodicity and consistency indices of behavioral patterns are calculated, thereby identifying the daily behavioral trajectory patterns of the elderly person, such as frequently used paths and fixed activity area sequences.

[0030] Furthermore, by utilizing the behavioral trajectory pattern, this embodiment of the invention constructs an activity space map of the elderly, which can structure the identified dynamic movement trajectories and static residence hotspots, and then clearly display the characteristics of the elderly's living space usage in a visual manner.

[0031] The activity space map refers to a data model that can intuitively reflect the characteristics of elderly people's use of living space and the logic of their activities, including information such as frequency of stay, duration of stay, and distribution of movement paths.

[0032] As an embodiment of the present invention, the activity space map of the elderly is constructed using the behavioral trajectory pattern, including: Identify high-frequency static hotspots and mainstream dynamic movement paths in the behavioral trajectory patterns; Using the aforementioned mainstream dynamic movement paths, a spatial activity network model for the elderly is constructed. The high-frequency static hotspots are spatiotemporally superimposed with the spatial activity network model to obtain an activity space map.

[0033] The high-frequency static dwelling hotspots refer to fixed locations where elderly people repeatedly appear within a specific time period (such as a single day or a week) and the duration of each stay exceeds a preset threshold. For example, if an elderly person consistently stays in the living room sofa area for more than 30 minutes between 2 pm and 4 pm for several consecutive days, this area can be identified as a high-frequency static dwelling hotspot. The mainstream dynamic movement paths refer to the spatial routes that elderly people frequently use and that have regularity. These paths have a high repetition rate in the movement sequence and their directions are relatively stable. The spatial activity network model refers to a network structure that describes the spatial activity patterns of elderly people and their regional correlations.

[0034] In detail, when identifying high-frequency static hotspots and mainstream dynamic movement paths, frequency statistical analysis of historical behavioral trajectory data can be performed to automatically filter out static hotspots whose frequency exceeds a set threshold (e.g., staying at a certain point for more than 30 minutes and more than 3 times; the specific threshold needs to be set based on the actual application data). Simultaneously, the main travel paths are extracted based on the recurring patterns of movement trajectories. A graph theory modeling method is used, treating each path node as a network vertex and the connectivity between nodes as network edges. By calculating the travel frequency and directional characteristics of each edge, a spatial network model with weighted attributes is established. Spatial registration technology is used to map hotspot areas to the corresponding node positions in the network model, and nodes are weighted according to the dwell time attribute of the hotspots. Based on this, combined with the weight characteristics of the edges in the network model, an activity space map that reflects both spatial distribution and behavioral characteristics is generated, where node size represents dwell time and edge thickness represents travel frequency.

[0035] Furthermore, by analyzing the activity patterns of the elderly based on the activity space map, this embodiment of the invention can identify stable activity patterns of the elderly in their living space. For example, the map analysis can reveal that the elderly always go to the balcony at 10 a.m. every day and travel to and from the study three times a week, thus intuitively presenting the stability and preferences of their spatial activities.

[0036] As an embodiment of the present invention, analyzing the activity pattern characteristics of the elderly based on the activity space map includes: Query the historical activity data of the elderly person to construct a baseline activity map of the elderly person using the historical activity data; Extract spatial distribution indicators of the same dimension from the activity space map and the benchmark activity map to obtain a comparison indicator set; Based on the comparison index set, the activity space map is annotated with abnormal regions to obtain annotated activity regions; The activity patterns of the elderly can be identified by using the marked activity areas.

[0037] The baseline activity map refers to a standardized spatial distribution map reflecting the normal activity patterns of the elderly, constructed based on long-term historical activity data. The comparison index set refers to a set of spatial distribution parameters extracted from the current activity spatial map and the baseline activity map for quantitative comparison. The marked activity area refers to the spatial range in the activity spatial map that is identified and specially marked by the system and has a significant deviation from the baseline pattern.

[0038] In detail, historical activity data stored in local edge nodes can be retrieved first, and then a baseline activity map for the elderly can be constructed using the method of constructing an activity spatial map. The same spatial distribution indicators are selected from the current activity spatial map and the baseline activity map for comparative analysis. These indicators mainly include key parameters such as activity area coverage, hotspot distribution density, and inter-regional movement frequency. By calculating the numerical differences of each indicator in the current map and the baseline map, a systematic set of comparison indicators is formed. When marking abnormal areas based on the comparison indicator set, a reasonable deviation threshold can be set first (such as the activity area exceeding the baseline by 10%, the dwell time exceeding the baseline by 10%, etc.). When the difference between the activity indicator of a certain area and the baseline value exceeds this threshold, it is determined that there is an activity abnormality in that area. Pattern analysis is performed on all marked activity areas to statistically analyze the spatial distribution patterns, occurrence time patterns, and duration characteristics of abnormal areas. By analyzing the combination patterns of different types of abnormal areas, the changing trends of the elderly's activity patterns can be identified, including patterns with early warning value such as shrinking activity range, weakened activity intensity, or changes in activity paths.

[0039] S3. Analyze the device usage characteristics of the elderly based on the device interaction signals, construct the habitual behavior chain of the elderly based on the device usage characteristics, and analyze the daily routine characteristics of the elderly based on the habitual behavior chain.

[0040] This invention, through analyzing the device usage characteristics of the elderly based on the device interaction signals, can extract the usage patterns of the elderly from the original device interaction signals. For example, it can analyze the time the elderly use the smart rice cooker every morning and evening, the frequency of triggering the smart door lock every week, etc., to clearly present their device operation preferences and habits.

[0041] As an embodiment of the present invention, analyzing the device usage characteristics of the elderly person based on the device interaction signals includes: Using the device interaction signals, a device usage table for the elderly is constructed, which includes device usage events and usage timestamps; Based on the device usage table, the frequency of use and average usage time of each device used by the elderly were analyzed; Based on the frequency of use and the average duration of use, the device usage characteristics of the elderly are identified.

[0042] In detail, the process begins by identifying the device identifier, state change type, and corresponding time information contained in the device interaction signals. Then, a standardized data table structure is established to record each device usage event in chronological order, forming a device usage table containing key fields such as device type, operation action, and timestamp. The usage frequency is calculated by counting the number of times each device is started within a set statistical period. Simultaneously, the average usage time of the same device is obtained by summing the duration of each use and dividing by the number of uses. Combining the usage frequency and average usage time of each device, devices are classified by setting frequency thresholds (e.g., values ​​above a certain number indicate high-frequency use) and duration thresholds (e.g., values ​​longer than a certain number indicate long-term use). For example, this identifies smart lights that are used frequently but for short periods, and televisions that are used infrequently but for long periods. At the same time, the usage timestamps are correlated to analyze the time period distribution, such as rice cookers being used mainly in the morning, noon, and evening. Finally, these are integrated to form device usage characteristics that include device preferences, usage intensity, and time period patterns.

[0043] Furthermore, in this embodiment of the invention, by constructing the habitual behavior chain of the elderly based on the device usage characteristics, the scattered device usage characteristics can be linked together by time or logic to form a sequence reflecting the coherent behavioral habits of the elderly.

[0044] The habitual behavior chain refers to a series of coherent and orderly behavioral sequences that an individual automatically executes without conscious or minimal conscious involvement, reflecting the device usage sequence combination of the elderly's daily fixed behavioral patterns.

[0045] As an embodiment of the present invention, based on the device usage characteristics, a habitual behavior chain of the elderly is constructed, including: The device usage events in the device usage characteristics are sorted by timestamps to obtain a device usage event sequence; The device usage event sequence is subjected to association rule mining to identify device usage events that are frequently and continuously triggered in time sequence, thereby obtaining frequent sequence patterns; Based on the frequent sequence patterns, the habitual behavior chain of the elderly is constructed.

[0046] In detail, the collected device usage events can be sorted in ascending order by timestamp to form a complete timeline sequence, i.e., a device usage event sequence. When identifying frequent sequence patterns, a sequence pattern mining algorithm can be used to analyze the device usage event sequence, focusing on discovering device usage combinations that are frequently and continuously triggered within a set time window (such as frequent use of fitness equipment within 1 hour). By setting a reasonable minimum support threshold, device usage sequences that repeat a certain number of times within the statistical period can be filtered out. The Apriori sequence pattern mining algorithm is used to scan the entire event sequence and count the frequency of specific event combinations (such as "turning on the smart kettle" followed by "using the smart pillbox") within a specified time window. When the frequency of an event sequence exceeds a preset threshold (which can be learned from historical data statistics), the sequence is determined to be a frequent pattern.

[0047] Furthermore, by analyzing the daily routine characteristics of the elderly based on the habitual behavior chain, the embodiments of the present invention can transform the elderly’s daily device usage behavior sequence into quantitative indicators reflecting their personal life rhythm and stability, thereby helping users to discover the elderly’s irregular behavior at a certain time or period.

[0048] As an embodiment of the present invention, the daily routine characteristics of the elderly are analyzed based on the habitual behavior chain, including: The habitual behavior chain is filtered for key time periods to obtain the core behavior time period group; Based on the core behavior time period groups, identify the time distribution pattern of the habitual behavior chain; By utilizing the aforementioned time distribution patterns, key behavioral periods of the elderly can be identified, thus revealing critical daily routines. By utilizing the aforementioned key sleep periods, the sleep patterns of the elderly can be identified.

[0049] In detail, the usage events of each device in the habitual behavior chain can be time-stamped, and then key behavioral nodes strongly related to daily life (such as turning on smart lights associated with waking up, turning off the TV associated with going to bed, etc.) can be filtered out to obtain core behavior time periods. Statistical analysis is then performed on the identified core behavior time periods to calculate the mean, variance, and distribution patterns of behavioral events within each time period group. By analyzing the consistency of behavioral event occurrence times within the same time period group on different dates, the time distribution patterns of the elderly in each core behavior time period can be identified, including the stability and fluctuation range of time points. Based on time... Statistical analysis of distribution patterns identifies core behavioral periods with high time stability and frequency meeting preset standards as key daily routine periods. Reasonable time fluctuation and frequency thresholds ensure sufficient representativeness and reliability of these periods, accurately reflecting the fixed daily routines of the elderly. These thresholds can be set in conjunction with historical data on normal elderly behavior. The stability of the start and end times, durations, and intervals of all key daily routine periods are considered as characteristics of daily routine patterns, such as "routine type," "routine regularity score," and "potential risk indicator."

[0050] S4. Using the activity pattern characteristics and the daily routine characteristics, analyze the behavioral deviation degree of the elderly's daily behavior, and construct a behavioral risk warning report for the elderly based on the behavioral deviation degree.

[0051] This invention utilizes the activity pattern characteristics and the daily routine characteristics to analyze the deviation of the elderly's daily behavior. By comparing the elderly's actual behavior with the established activity and daily routines, the degree of behavioral deviation can be quantified. For example, if an elderly person usually gets up at 7 a.m. to turn on the lights and goes to the balcony at 10 a.m., but now gets up at 9 a.m. and does not go to the balcony, the deviation of this behavior from the established routine can be accurately analyzed, and the abnormal behavior can be presented intuitively.

[0052] As an embodiment of the present invention, the behavioral deviation of the daily behavior of the elderly is analyzed using the activity pattern characteristics and the daily routine characteristics, including: Using the aforementioned activity patterns, the intensity of abnormal fluctuations in the daily behavior of the elderly over time was analyzed. The intensity of the fluctuation anomaly is used as the first deviation factor of the elderly person's behavior; Using the aforementioned characteristics of daily routines, the degree of disruption in the periodicity and stability of the daily routines of the elderly was analyzed, resulting in a second deviation factor; Using the first deviation factor and the second deviation factor, calculate the comprehensive behavioral deviation coefficient of the elderly person's daily behavior; Based on the comprehensive behavioral deviation coefficient, the behavioral deviation degree of the elderly person's daily behavior is determined.

[0053] The first deviation factor refers to an indicator that quantifies the deviation of an elderly person from their personal historical baseline in terms of spatial activity range and movement frequency. The degree of disruption refers to the degree of interruption and disorder in the elderly person's daily routine. The second deviation factor refers to an indicator that quantifies the deviation of an elderly person from their personal historical baseline in terms of daily routine time, sleep cycle, and daily routine stability.

[0054] In detail, time series anomaly detection algorithms (such as Z-score-based statistical process control) can be used. Key indicators from activity pattern characteristics, such as activity frequency, duration, and spatial movement range, are used as input to calculate their mean and standard deviation within a sliding time window. When the deviation of the real-time collected data point from the individual's normal activity baseline established based on historical data exceeds a preset fluctuation threshold (e.g., the deviation in activity duration exceeds 1.5 times the standard deviation of the historical average), it is determined that there is an anomaly at that time point. The calculated anomaly intensity value is directly defined as the first deviation factor. This factor is a continuous variable; the larger its value, the greater the deviation in the elderly person's daily activity pattern relative to their normal activity pattern over time. The more significant the perturbation of the personal baseline, the more effective the use of time-series pattern matching techniques (such as Dynamic Time Warping (DTW) or periodic analysis) to compare the sequence of daily routine events (such as key times for waking up, eating, and going to bed) in the current cycle (such as the last three days) with a personal standard daily routine established based on historical data. By calculating the time distance or phase shift between the two, the degree of periodic disruption of the daily routine can be quantified. When determining the degree of behavioral deviation based on the comprehensive behavioral deviation coefficient, a three-level threshold can be set: when the comprehensive coefficient is below 0.3, the behavioral deviation is judged as "normal" level; when the coefficient is between 0.3 and 0.7, it is judged as "concern" level; and when the coefficient is above 0.7, it is judged as "abnormal" level, indicating high risk.

[0055] Preferably, the comprehensive behavioral deviation coefficient of the elderly person's daily behavior is calculated using the first deviation factor and the second deviation factor, including: The first deviation factor and the second deviation factor are nonlinearly corrected to obtain the first linear factor and the second linear factor. The first linear factor and the second linear factor are adjusted for differences to obtain the first enhancement factor and the second enhancement factor. Based on the first enhancement factor and the second enhancement factor, the comprehensive behavioral deviation coefficient of the elderly person's daily behavior is calculated.

[0056] In detail, the first deviation factor can be nonlinearly corrected using the following formula:

[0057] in, Indicates the first linear factor. Indicates the first deviation factor. This represents the first nonlinear adjustment coefficient; The second deviation factor can be nonlinearly corrected using the following formula:

[0058] in, Indicates the second linear factor. This represents the second deviation factor. This represents the second nonlinear adjustment coefficient.

[0059] Furthermore, the difference correction for the first linear factor can be performed using the following formula:

[0060] in, Indicates the first enhancing factor. Indicates the first linear factor. , Indicates the index of increased difference; The difference correction for the second linear factor is performed using the following formula:

[0061] in, Indicates the second enhancing factor. Indicates the first linear factor. , This indicates the difference enhancement index.

[0062] Furthermore, the overall behavioral deviation coefficient can be calculated using the following formula:

[0063] in, This represents the overall behavioral deviation coefficient. Indicates the first enhancing factor. Indicates the second enhancing factor. Indicates and The corresponding weighting coefficients, Indicates and The corresponding weighting coefficients.

[0064] It should be further explained that the nonlinear adjustment coefficients for nonlinear correction of the first deviation factor and the second deviation factor are determined by constructing a sample set through collecting behavioral data of elderly people in normal and known abnormal states, with deviation judgment accuracy and recall as optimization objectives, and then through cross-validation. The specific values ​​need to be calculated and determined in conjunction with the actual sample data. The difference enhancement index α is a unified global correction coefficient used to control the difference enhancement magnitude between the first enhancement factor and the second enhancement factor. Using the same difference enhancement index can ensure the symmetry of the first deviation factor and the second deviation factor in the difference correction process. The difference enhancement index can be obtained by fitting historical behavioral data. The specific values ​​need to be calculated and determined in conjunction with the actual sample data.

[0065] Furthermore, this embodiment of the invention can construct a behavioral risk warning report for the elderly based on the behavioral deviation, which can transform the quantified behavioral deviation into a specific report that includes the deviation scenario and risk level, and intuitively present the risk situation of abnormal behavior in the elderly. For example, when the elderly’s wake-up time deviates from the norm by 2 hours and the medication reminder device is not operated, the generated warning report can provide caregivers with a clear focus and timely grasp their potential risks.

[0066] The aforementioned behavioral risk warning report refers to a numerical indicator that reflects the degree of abnormal behavior in the elderly.

[0067] As an embodiment of the present invention, a behavioral risk warning report for the elderly is constructed based on the behavioral deviation degree, including: Based on the behavioral deviation, a behavioral risk level for the elderly is generated; Based on the behavioral risk level, construct a behavioral risk early warning report for the elderly.

[0068] The behavioral risk level refers to a structured early warning document automatically generated based on a determined risk level.

[0069] In detail, the behavioral risk level can be determined based on the calculation results of the behavioral deviation. For example, a deviation between 0.1 and 0.3 generates a low risk level, a deviation between 0.3 and 0.5 generates a medium risk level, and a deviation greater than 0.5 generates a high risk level. The specific classification can be further refined according to actual application needs. Finally, the behavioral characteristic data of the elderly, the behavioral deviation corresponding to different characteristics, and the behavioral risk level corresponding to different behavioral deviations are compiled into a table to construct an early warning report on behavioral risks of the elderly.

[0070] like Figure 2 The diagram shown is a functional block diagram of the elderly daily behavior pattern analysis and risk warning system based on edge computing of the present invention.

[0071] The edge computing-based daily behavior pattern analysis and risk warning system 200 for the elderly described in this invention can be installed in an electronic device. Depending on the functions implemented, the edge computing-based daily behavior pattern analysis and risk warning system for the elderly may include a behavior signal acquisition module 201, an activity pattern analysis module 202, a daily routine analysis module 203, and an anomaly warning module 204. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0072] In this embodiment of the invention, the functions of each module / unit are as follows: The behavior signal acquisition module 201 is used to continuously collect multi-source behavior data of the elderly at the edge computing node deployed in the elderly's living space. The multi-source behavior data includes spatial positioning signals and device interaction signals. The activity pattern analysis module 202 is used to identify the behavioral trajectory patterns of the elderly using the spatial positioning signal. The behavioral trajectory patterns include dynamic movement trajectories and static hotspots. The module uses the behavioral trajectory patterns to construct the activity space map of the elderly and analyzes the activity pattern characteristics of the elderly based on the activity space map. The daily routine analysis module 203 is used to analyze the device usage characteristics of the elderly based on the device interaction signals, construct the habitual behavior chain of the elderly based on the device usage characteristics, and analyze the daily routine characteristics of the elderly based on the habitual behavior chain. The abnormal early warning module 204 is used to analyze the behavioral deviation of the elderly's daily behavior by utilizing the activity pattern characteristics and the daily routine characteristics, and to construct a behavioral risk early warning report for the elderly based on the behavioral deviation.

[0073] In detail, the modules in the edge computing-based elderly daily behavior pattern analysis and risk warning system 200 described in this embodiment of the invention employ the same methods as described above. Figure 1 The method used is the same as the edge computing-based analysis and risk warning method for the daily behavior patterns of the elderly described in the article, and can produce the same technical effect, so it will not be elaborated here.

[0074] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0075] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing, characterized in that, The method includes: At edge computing nodes deployed in the living spaces of the elderly, multi-source behavioral data of the elderly are continuously collected. The multi-source behavioral data includes spatial positioning signals and device interaction signals. Using the spatial positioning signal, the behavioral trajectory pattern of the elderly is identified. The behavioral trajectory pattern includes dynamic movement trajectory and static stationary hotspot. Using the behavioral trajectory pattern, an activity space map of the elderly is constructed. Based on the activity space map, the activity pattern characteristics of the elderly are analyzed. Based on the device interaction signals, analyze the device usage characteristics of the elderly, construct the habitual behavior chain of the elderly based on the device usage characteristics, and analyze the daily routine characteristics of the elderly based on the habitual behavior chain. By utilizing the activity pattern characteristics and the daily routine characteristics, the behavioral deviation degree of the elderly is analyzed, and a behavioral risk warning report for the elderly is constructed based on the behavioral deviation degree.

2. The method for analyzing and predicting risks of daily behavior patterns of the elderly based on edge computing as described in claim 1, characterized in that, Using the aforementioned activity pattern characteristics and daily routine characteristics, the behavioral deviation of the elderly's daily behavior is analyzed, including: Using the aforementioned activity patterns, the intensity of abnormal fluctuations in the daily behavior of the elderly over time was analyzed. The intensity of the fluctuation anomaly is used as the first deviation factor of the elderly person's behavior; Using the aforementioned characteristics of daily routines, the degree of disruption in the periodicity and stability of the daily routines of the elderly was analyzed, resulting in a second deviation factor; Using the first deviation factor and the second deviation factor, calculate the comprehensive behavioral deviation coefficient of the elderly person's daily behavior; Based on the comprehensive behavioral deviation coefficient, the behavioral deviation degree of the elderly person's daily behavior is determined.

3. The method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing as described in claim 2, characterized in that, Using the first deviation factor and the second deviation factor, the comprehensive behavioral deviation coefficient of the elderly person's daily behavior is calculated, including: The first deviation factor and the second deviation factor are nonlinearly corrected to obtain the first linear factor and the second linear factor. The first linear factor and the second linear factor are adjusted for differences to obtain the first enhancement factor and the second enhancement factor. Based on the first enhancement factor and the second enhancement factor, the comprehensive behavioral deviation coefficient of the elderly person's daily behavior is calculated.

4. The method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing as described in claim 1, utilizing the spatial positioning signal to identify the behavioral trajectory patterns of the elderly, includes: Trajectory point clustering analysis is performed on the spatial positioning signal to identify the static hotspots and dwell time of the elderly using the decomposition results of the trajectory point clustering analysis; Identify the spatial positioning sequence of the spatial positioning signal, and use the spatial positioning sequence to construct the dynamic movement trajectory of the elderly person; Based on the static hotspots, the duration of stay, and the dynamic movement trajectory, the behavioral trajectory patterns of the elderly are identified.

5. The method for analyzing and risk warning of daily behavior patterns of the elderly based on edge computing as described in claim 1, wherein the activity pattern characteristics of the elderly are analyzed according to the activity space map, including: Query the historical activity data of the elderly person to construct a baseline activity map of the elderly person using the historical activity data; Extract spatial distribution indicators of the same dimension from the activity space map and the benchmark activity map to obtain a comparison indicator set; Based on the comparison index set, the activity space map is annotated with abnormal regions to obtain annotated activity regions; The activity patterns of the elderly can be identified by using the marked activity areas.

6. The method for analyzing and predicting risks of daily behavior patterns of the elderly based on edge computing as described in claim 1, characterized in that, Based on the aforementioned habitual behavior chain, the daily routine characteristics of the elderly were analyzed, including: The habitual behavior chain is filtered for key time periods to obtain the core behavior time period group; Based on the core behavior time period groups, identify the time distribution pattern of the habitual behavior chain; By utilizing the aforementioned time distribution patterns, key behavioral periods of the elderly can be identified, thus revealing critical daily routines. By utilizing the aforementioned key sleep periods, the sleep patterns of the elderly can be identified.

7. The method for analyzing and predicting risks of daily behavior patterns of the elderly based on edge computing as described in claim 1, characterized in that, Based on the device interaction signals, the device usage characteristics of the elderly person are analyzed, including: Using the device interaction signals, a device usage table for the elderly is constructed, which includes device usage events and usage timestamps; Based on the device usage table, the frequency of use and average usage time of each device used by the elderly were analyzed; Based on the frequency of use and the average duration of use, the device usage characteristics of the elderly are identified.

8. The method for analyzing and predicting risks of daily behavior patterns of the elderly based on edge computing as described in claim 1, characterized in that, Using the aforementioned behavioral trajectory patterns, an activity space map of the elderly person is constructed, including: Identify high-frequency static hotspots and mainstream dynamic movement paths in the behavioral trajectory patterns; Using the aforementioned mainstream dynamic movement paths, a spatial activity network model for the elderly is constructed. The high-frequency static hotspots are spatiotemporally superimposed with the spatial activity network model to obtain an activity space map.

9. The method for analyzing and predicting risks of daily behavior patterns of the elderly based on edge computing as described in claim 1, characterized in that, Based on the device usage characteristics, a habitual behavior chain of the elderly is constructed, including: The device usage events in the device usage characteristics are sorted by timestamps to obtain a device usage event sequence; The device usage event sequence is subjected to association rule mining to identify device usage events that are frequently and continuously triggered in time sequence, thereby obtaining frequent sequence patterns; Based on the frequent sequence patterns, the habitual behavior chain of the elderly is constructed.

10. The method for analyzing and predicting the daily behavior patterns of the elderly based on edge computing as described in claim 1, characterized in that, Based on the behavioral deviation, a behavioral risk warning report for the elderly is constructed, including: Based on the behavioral deviation, a behavioral risk level for the elderly is generated; Based on the behavioral risk level, construct a behavioral risk early warning report for the elderly.

Images